Improving schema understandability for semantic integration

点击翻译按钮获取中文摘要

Databases are very complex; a typical database can store data in hundreds of tables and can contain hundreds of references between tables. The layout of the data into tables and references between them is referred to as a schema. Learning the details of these complex schemas requires significant time. This problem is exacerbated by data sources being shared across organizations, and users attempting to make their applications do something for which they were not originally designed. Consider the following two scenarios: (1) A user is integrating civil engineering data across a number of applications, including CAD models and project scheduling. The data in these applications can be exported via a number of different standards. In order to integrate the data, the user must first choose which standard to export from and then figure out how to understand how the data has been exported. (2) An end user has access to a database application that generates reports. The end user asks the database programmer to change the application to access more information than was originally accessible. The programmer has never seen the application before, but now must extend the application to access new data. This will require the programmer to understand how the data in the back-end database is stored well enough to write the new queries. Neither of these scenarios is adequately helped by existing approaches. I propose to improve schema understanding. The goal is to help users in a variety of scenarios including those above to make better use of the existing schemas that they are encountering for the first time, whether it is extending their understanding of a current application, trying to decide how to export their data to a standard schema (which is challenging since related data may be scattered across the standard schema, rendering it incomprehensible), or choosing a standard schema to which the data should adhere.